The Impact of Trade- Strength on

over the Windward Side of the Island of

Mary Ann Esteban and Yi-Leng Chen1

Department of

School of Ocean and Science and Technology

University of Hawaii

Honolulu, HI 96822

June 2007

1Corresponding author address: Dr. Yi-Leng Chen, Department of Meteorology, SOEST,

University of Hawaii at Manoa, Honolulu, HI 96822, E-mail: [email protected] ABSTRACT

The effects of trade-wind strength and the diurnal heating cycle on the production of

summer trade-wind rainfall on the windward side of the island of Hawaii are examined from the

data collected from the Hawaiian Rainband Project (HaRP) during 11 July - 24 August 1990 and

the Hydronet and National Climatic Data Center gauge data

during 11 July - 24 August for years 1997-2000.

For strong trades, the daily rainfall totals on the windward lowlands west of Hilo are

higher with a nocturnal maximum there due to the convergence of the katabatic flow and the

incoming decelerating trade-wind flow, and orographic lifting aloft. The maximum rainfall axis

shifts farther inland when trades are stronger. Except in the late afternoon hours, rainfall

amounts on the windward side are higher when trades are stronger. For weak trades (≤ 5 m s-1), the rainfall distributions have a pronounced late afternoon maximum on the windward slopes due to the development of anabatic . The nocturnal rainfall over the windward lowlands and the early morning coastal rainfall are lower when trades are weaker. 1. Introduction

The island of Hawaii (Big Island) is the largest of the Hawaiian chain. The terrain of the

Big Island is dominated by two volcanic mountains, Mauna Loa and , both exceeding

4 km in height, well above the typical height of trade-wind inversion (~ 2 km). During the

month of August, trade winds blow 92% of the time, the highest frequency throughout the year

for the state (Sanderson 1993), because of the presence of the semi-permanent Pacific High

located northeast of the island chain.

Prior to HaRP (Hawaiian Rainband Project, 11 July – 24 August, 1990), modeling and

observational studies by Smolarkiewicz et al. (1988) and Rasmussen et al. (1989) suggested that

the formation of the Hilo cloud band results from the dynamic interaction of the trade winds with

the island. They theorize that the low-level winds on the windward side of the island reverse

direction under a small (< 1) Froude number (Fr) flow regime as a result of orographic blocking.

The Froude number is given by Fr = U/NH, where U is the mean wind speed, N is the Brünt-

Väisälä frequency, and H is the height of the obstacle. A cloud band forms where dynamically

induced reverse flow and the incoming trades converge. Comparing pressure perturbations and

surface airflow between the composited 12 strongest (Fr ~ 0.3) and 12 weakest (Fr ~ 0.2) trade-

wind days during HaRP, Chen and Nash (1994) showed that when upstream trades are stronger,

the orographically induced high-pressure cell on the windward side (Smolarkiewicz et al. 1988)

is higher (0.2-0.5 hPa) with lower pressure in the lee-side wake zone.

Rasmussen and Smolarkiewicz (1993) showed that, in their model, the nocturnal cooling

doubles the depth and speed of the reverse flow. However, they maintain that the convergence

line offshore is primarily a result of dynamic forcing; the nocturnal cooling has very little effect on the position of the convergence line. However, it is important to note that the effects of

3 nocturnal cooling are underestimated by their model (Feng and Chen, 2001). In the early

morning before sunrise, the virtual temperature differences between the surface air on the

windward side and the upstream value at the same altitude reach -4 K (Chen and Wang 1994;

Feng and Chen 1998) and are only ~ -1 K in the modeling studies of Smolarkiewicz et al. (1988)

and Rasmussen et al. (1989).

Using PAM (Portable Automatic Mesonet) surface data collected during HaRP, Chen and

Nash (1994) and Chen and Wang (1994) studied the precipitation, airflow, and thermodynamic

fields over the island of Hawaii throughout the diurnal cycle. Chen and Nash (1994) showed that

with mean weak winds caused by orographic blocking on the windward side, the thermally

driven circulations there become significant. Chen and Wang (1994) found that, on the

windward side, the onset of anabatic (katabatic) flow occurs after sunrise (before sunset) on the

upper slopes where virtual temperature first becomes warmer (cooler) than environmental values.

The timing of the wind shift during these periods can also be affected by rainshowers and cloud

cover that modify the thermodynamic fields near the surface (Chen and Wang 1995). Carbone et

al. (1995) show that the leading edge of the downslope flow resembles a density current.

Throughout the night, the katabatic flow deepens and intensifies, then gradually extends over the

ocean (Feng and Chen 1998). The location of the between the offshore flow

and the incoming trades is related to the offshore extension of the katabatic flow (Feng and Chen

1998) and is not only dependent on Fr as suggested by modeling studies (Smolarkiewicz et al.

1988; Rasmussen and Smolarkiewicz 1993). Rain evaporative cooling serves to deepen the

katabatic flow (Wang and Chen 1995). The strength of the upstream trades affects the onset time

of the katabatic (anabatic)/offshore (onshore) flow in the late afternoon (early morning). Under

4 stronger trade winds, offshore flow over the windward side begins later in the evening and ends

earlier in the morning (Chen and Nash 1994).

Reisner and Smolarkiewicz (1994) studied low (< 1) Fr flow past a three-dimensional obstacle with uniform heating at the surface and obtained a simple criterion for the transition from the blocked flow to an unblocked flow. Raymond and Rotunno (1989) and Lin et al. (1993) introduced thermal Froude number, U/Nd, where d is the depth of the cooling/heating region, to study flow response forced by prescribed heat sinks when the environment is stably stratified. In the work of Lin et al. (1993), the prescribed cooling region is set at the center of their 2-D model

domain with a depth of 3 km. A density current is able to form only when the cooling is strong enough to produce an outflow that spreads faster than the compensating gravity waves. Even though the leading edge of land breezes on the windward side of the island of Hawaii resembles a density current, to study the land/sea-breeze circulations, mountain/valley winds and the secondary circulations forced by land-sea thermally contrast for the island of Hawaii throughout the diurnal cycle requires a 3-D mesoscale model. Furthermore, the results from 2-D simulations may not be applicable to 3-D flows as flow splitting occurs on the windward side of the island of

Hawaii owing to the fact that the mountain height are well above the trade-wind inversion

(Leopold 1949; Schär and Smith 1993). The incoming flow may be deflected around the obstacle rather than passing though the heat sink as in a 2-D model.

Chen and Feng (2001) use the composite upstream sounding derived from 20 HaRP

soundings taken by the NCAR Electra to initialize the Pennsylvania State University-National

Center for Atmospheric Research (PSU-NCAR) fifth generation Mesoscale Model (MM5;

Dudhia 1993) without considering the diurnal heating cycle. They show that in addition to Fr,

the simulated island airflow and cloud distributions are sensitive to trade-wind inversion height

5 (Chen and Feng 1995) and net diabatic heating rate associated with clouds and precipitation.

With nocturnal cooling, Feng and Chen (2001) successfully simulated the major observed features associated with the nocturnal flow regime, including the formation of nocturnal

inversion, development and gradual offshore extension and deepening of the katabatic flow,

shifting of the overall cloudy area from the windward slopes to the ocean, and generation of

clouds within the offshore convergence zone in the early morning. They also simulated the

dependence of depth, strength, and offshore extension of the offshore flow on rain evaporation.

Recently, the full diurnal cycles of island-scale airflow and weather over the island of Hawaii

were successfully simulated by Yang et al. (2005) using MM5 coupled with an advanced land

surface model (Chen and Dudhia 2001) with proper treatments of lower boundary conditions

over land (Zhang et al. 2005). In their model, the thermally direct diurnal circulation cells on

the windward side for both the daytime and night flow regimes are reproduced (Yang et al. 2005;

figs. 25 and 26).

HaRP studies reveal that there are three rainfall regimes during the diurnal cycle on the

windward side: the nocturnal rainfall regime over the windward lowlands, the early morning

coastal rainfall regime, and the daytime orographic rainfall regime (Chen and Nash 1994; Chen

and Feng 1995). During the evening hours, rainfall totals are at their maximum (Schroeder et al.

1977) over the windward lowlands as a result of the convergence between the katabatic flow and

the moist incoming trade winds inland (Chen and Nash 1994; Feng and Chen 1998). The rising

motion caused by the low-level convergence is enhanced by orographic lifting aloft (Feng and

Chen 2001). With further cooling, the convergence zone and the maximum rainfall axis shift

toward the coast as the leading edge of the katabatic flow propagates toward the coast and

deepens (Chen and Nash 1994; Feng and Chen 1998). In the early morning, rainbands are

6 frequently observed offshore in the convergence zone (Leopold 1949; Lavoie 1967; Garrett

1980). The origin of nocturnal and early morning rainfall may be attributed to pre-existing upstream cloud patches (Larson 1978; Austin et al. 1996; Wang and Chen 1998; Wang et al.

2000) that are enhanced by the flow deceleration and the convergence zone (Rasmussen et al.

1993; Carbone et al. 1998; Wang et al., 2000) or to rainshowers that form within the convergence zone (Leopold 1949; Takahashi 1981; Rasmussen et al. 1993; Lee and Chen 1999).

In the afternoon hours, orographic clouds develop on the slopes as a result of the development of combined anabatic and sea breeze winds (Leopold 1949; Lavoie 1967; Garrett 1980; Feng and

Chen 1998) with relatively clear skies over the ocean. The diurnal rainfall regimes over the island of Hawaii are well simulated by our recent modeling studies (Yang et al. 2005).

The main objective of this study is to improve our understanding of rainfall production on the windward side of the island of Hawaii. We will study the diurnal rainfall variations on the windward side under various trade-wind strengths to investigate the effects of trade-wind strength on the rainfall production for different diurnal rainfall regimes. Rainfall diurnal cycles for the 12 strongest and 12 weakest trade-wind days during the HaRP period, characterized by

Chen and Nash (1994) will be analyzed. In addition, we would like to extend our analysis of rainfall diurnal cycle under various trade-wind strengths to the summers of 1997-2000 using the routine rain gauge data and the rainfall data from the newly deployed hydronet over the

Hawaiian Islands.

Extending the analysis to 1997-2000 will serve two purposes. The first is to expand the database to more samples in order to further eliminate uncertainties and biases. Secondly, this will help determine whether the rainfall patterns from the HaRP experiment in 1990 are

7 representative or not. Other features on a synoptic scale that may affect the trade-wind strength

over the Hawaiian Islands will be discussed.

2. Data Sources

Data from the 50 NCAR PAM stations during HaRP (Fig. 1) are used. Pressure,

temperature, wet-bulb temperature, u- and v-wind components, rainfall and radiation

measurements* (from stations 13, 15, and 16) are recorded at one-minute intervals. Surface

wind fields are based on 15-minute averages as described in Chen and Nash (1994). Rainfall

measurements at each PAM station have a resolution of 0.254 mm. Station 15 contains faulty

rainfall data and the data from this station are not used in the analysis.

Out of the 45 HaRP days, the 12 strongest and 12 weakest trade-wind days were

determined from wind data at PAM stations from the northernmost (Hawi) and southernmost

(South Point) locations by Chen and Nash (1994). Winds averaged 7.8 m s-1 for the 12 strongest

days, which are July 11, 13, 24, 25, 28 and August 8, 9, 12, 13, 16, 18, 23. The weakest trade-

wind days are July 15, 20, 21, 27, 29, 30 and August 3, 5, 7, 19, 20, 21, with an average wind

speed of 5.6 m s-1.

The classification of trade-wind strength from the 1990 HaRP wind data is consistent

with the NCEP (National Centers for Environmental Prediction)/NCAR’s 1990 reanalysis wind

data taken from the grid point 200N, 1550W at the 1000-hPa level, except the wind speed from

the NCEP/NCAR reanalysis is slightly lower. The NCEP/NCAR reanalysis may underestimate the wind speed over the open ocean immediately upstream of Hilo (Yang et al. 2005) but still can

be used to rank the trade-wind strength during the period. The averaged wind speed and the

______

*Radiation measurements were only taken at stations 13, 15, and 16.

8 standard deviations from the 1990 NCEP/NCAR reanalysis wind data for the 12 strong and the

12 weak HaRP days are shown in Table 1. Winds for the strong and weak trade days are

significantly different with a confidence level beyond 98% using the Kolgomorov-Smirnov (KS)

hypothesis test (Weiss and Hassett, 1991).

Expansion of the dataset to include the period of 11 July - 24 August for years 1997-2000

is made. The NCEP/NCAR reanalysis wind data at 1000-hPa level from the grid-point 200N,

1550W are used to rank the trade-wind strength. The years 1999 and 2000 contained significantly stronger trades as compared with HaRP. Days having a positive v-wind component are eliminated. A total of six wind categories are developed for 1997-2000 using the HaRP

mean daily wind speeds for the 12 strong (us ) and the 12 weak (uw ) trade-wind days and their

standard deviations (σ s and σ w) determined from the NCAR/NCEP reanalysis 1000-hPa wind

data at the same grid point as references (Table 2). The weakest trade-wind days are those with a

mean daily wind speed less than uw – σ w. The weak trade-wind days are those with a mean

daily wind speed between uw – σ w and uw + σ w. Moderate trade-wind days are those having a

mean daily wind speed ranging from uw + σ w to us – σ s. The strong trade-wind days are those

with a mean daily wind speed between us – σ s and us + σ s. The moderately strong trade-wind

days are those having a mean daily wind speed from us + σ s to us + 2σ s. The very strong

trade-wind days are those having a mean daily wind speed greater than us + 2σs. The KS

significance test results show that the winds in different trade-wind regimes differ significantly

with a confidence level beyond 98%.

For the very strong trade-wind regime, the semi-permanent subtropical high (~1026 hPa)

to the north of the island chain dominates the synoptic map, which features relatively large

9 pressure gradients over the Hawaiian Islands (Fig. 2a). In contrast, the weakest trade-wind

composite map (Fig. 2b) shows weaker pressure gradients because the high pressure cell is split

by the passage of a mid-latitude cold front farther to the north. Chen and Feng (1995) have

shown that for the 20 July 1990 case, a mid-latitude storm passes farther to the north of the

island chain and splits the subtropical high.

Average inversion base heights from Hilo soundings for different trade-wind regimes

ranged between 789 hPa and 808 hPa, except for the weakest trade regime which had a lower

average inversion height of 827 hPa (Table 3). The geopotential height for the weakest trade

regime was slightly higher (Fig. 2b) because the ridge axis of the subtropical high is pushed

slightly southward during the passage of a mid-latitude storm.

The rainfall data from eighteen National Weather Service (NWS) Honolulu Forecast

Office (HFO) hydronet stations over the island of Hawaii (Table 4 and Fig. 3a) during the period

of 11 July - 24 August for years 1997-2000 are used. Rainfall data from these stations are

measured by a telemetered tipping bucket rain gauge. This real-time rain gauge network collects

data at 15-minute intervals with a resolution of 0.254 mm (0.01 in.). Many gauges are located in

remote locations throughout the State of Hawaii and data quality is variable. For some periods,

15-minute interval data were not available. The data provided are given in running totals, so when rainfall reaches 99.99 inches, it will be automatically reset to 0 inches. Individual counters are manually reset, denoted with the letter “m”, if any work or maintenance is performed on that particular gauge, regardless of the reading at that time.

The National Climatic Data Center (NCDC) rainfall data for the island of Hawaii (Table

5) will be added to the NWS Hydronet data. There are nineteen NCDC stations (Fig. 3b)

10 measuring hourly precipitation with a resolution of a 2.54 mm (0.1 in.). The only exceptions

come from Hilo Airport and Lihue Airport where the resolution is 0.254 mm (0.01 in.).

3. Transect from the Upper Slope to the Coast

A transect from the upper slopes to the coastline west of Hilo is constructed from data

collected from PAM stations located at Kulani Prison (Station 11, elevation 1579 m), North

Kulani Road (Station 10, elevation 506 m), Hawaiian Acres (Station 4, elevation 202 m), and

Paradise Park (Station 8, elevation 17 m) during HaRP to show the diurnal cycles of rainfall and averaged surface winds (Fig. 4).

For both the 12 strong and 12 weak trade-wind days during HaRP, the surface airflow

along the transect is dominated by the daytime upslope-onshore flow and nighttime katabatic-

offshore flow (Figs. 5a and 6a). Note that for the strong trade-wind regime, the nighttime

katabatic flows at stations 10 and 11 have a more northerly component compared to the weak

trade-wind regime. For weak trade-wind days (Fig. 6a), the daytime upslope flow has a more

easterly wind component compared to strong trade-wind days (Fig. 5a). Chen and Nash (1994) found that the u-component difference, ustrong – uweak, between the strong and weak trade-wind days showed a small positive difference (> 1 m s-1) on the windward slopes for both the daytime and nighttime flow regimes. The difference indicates that the upslope flow during the day is slightly weaker and the nighttime katabatic flow slightly stronger for the strong than the weak trade-wind days, consistent with the observed higher pressure deviations on the windward side when trades are stronger (Chen and Nash 1994).

Both the strong and weak trade regimes display an evening rainfall regime after sunset

(Figs. 5b and 6b). However, the nocturnal rainfall on the windward lower slopes is more frequent with a higher rain rate for strong trade-wind days (1.2 m h-1) (Fig. 5b) than weak trade-

11 wind days (Fig. 6b). The maximum hourly nocturnal rainfall frequency exceeds 60% under

strong trades (Fig. 5c). In comparison, the precipitation is less frequent at night for weak trade-

wind days (40%) (Fig. 6c) with lower mean rain rates (0.4 -0.8 mm h-1), except a local rainfall maximum (1.2 mm h-1) during 0200-0300 HST over the coast (Fig. 6b). Some of the nocturnal rainshowers originated far upstream and were enhanced by flow deceleration upstream, the convergence between the katabatic flow and the incoming trade-wind flow, and orographic lifting aloft (Carbone et al. 1998; Frye and Chen 2001). In some cases, trade-wind showers were generated within the leading edge of the katabatic flow (Lee and Chen 1999). Modeling studies

(Feng and Chen 2001; Yang et al. 2005; Yang and Chen 2007) show that at night, the thermally driven circulation cell on the windward side is characterized by descending airflow immediately above the slope surface, a convergence zone between the decelerating incoming trade-wind flow and the katabatic flow, and rising motion aloft due to orographic lifting. It appears that the rising motions caused by the convergence between decelerating incoming trades and the katabatic flow west of Hilo, and the orographic lifting aloft (Chen and Nash 1994) are more significant when trades are stronger, resulting in higher nocturnal rainfall there. Recent modeling work by Yang and Chen (2007) suggests that neither dynamic forcing (without land surface forcing)

(Smolarkiewicz et al. 1988) nor land surface forcing (without terrain) alone is adequate to account for the rainfall production on the windward side. From their model sensitivity tests with different terrain heights, they show that the rainfall production on the windward side in the model is negatively correlated with variations in Froude number due to differences in model terrain heights. The rainfall production on the windward side is closely related to rising motions caused by nonlinear interactions among orographic lifting, orographic blocking, land surface processes and feedback effects from .

12 For both strong and weak trade-wind days, as the katabatic flow extends toward the coast

throughout the night, the rainfall rates on the windward lowlands decrease. For both regimes,

rainfall frequency along the windward coastal region and over the windward lowlands is relatively high (30-60%) before sunrise. Modeling studies (Feng and Chen 2001; Yang et al.

2005) show that the simulated flow deceleration of the incoming trade-wind flow on the windward side is most significant in the early morning when the katabatic flow is the deepest with a maximum horizontal extent (Chen and Feng 1998) in agreement with aircraft observations

(Esteban 2002). The early morning rainfall along the coast is frequently related to pre-existing trader-wind rainshowers upstream that are enhanced within the offshore convergence zone as they move toward the island of Hawaii (Austin et al. 1996; Wang and Chen 1998; Wang et al.

2000). These rainshowers weaken and dissipate as they move over the cooler and drier offshore flow (Wang and Chen 1998).

In contrast to the nocturnal flow regime, daytime (0800-1600 HST) rainfall amounts are

slightly lower (Fig. 5b) with lower rainfall frequencies (~ 20%) (Fig. 5c) when trades are stronger. This is consistent with a slightly weaker upslope flow for strong trade-wind days as compared to weak trade-wind days (Chen and Nash 1994). A local rainfall maximum (0.8-1.6 mm h-1) during the same time period occurs over the upper slopes under weak trades with higher

rainfall frequencies (~ 40%) (Fig. 6c). Under weak trades, the daytime rainfall starts on the

windward lowlands at Hawaiian Acres (Station 4, elevation 202 m) in the morning (Fig. 6b).

Rainfall increases to more than 1.2 mm h-1 a few hours later at the lower slopes at North Kulani

Road (Station 10) (Fig. 6b) as a result of the establishment of upslope flow (Fig. 6a). The

greatest intensity occurs at station 10 in the late afternoon hours for weak trade-wind days with hourly rainfall frequencies > 40%. The trade-wind inversion height is another factor that affects

13 the rainfall production on the windward side. Feng and Chen (1995) show that for higher trade-

wind days, the afternoon orographic clouds are closer to the summits and more extensive with

higher rainfall amounts. The mean trade-wind inversion height is slightly higher for the weak

trade-wind days (762 hPa) than for the strong trade-wind days (778 hPa) (Table 3). However,

the difference in trade-wind inversion height between the strong and weak trade-wind days

during HaRP is relatively small. During the summers of 1997-2000, the mean trade-wind

inversion height is the lowest (827 hPa), when the trade-wind strength is the weakest (Table 3).

The impact of trade-wind strength on daytime rainfall on the windward side will be investigated

further in the next section.

The above results show that the effects of trade-wind speed upon windward rainfall

production vary throughout the diurnal cycle. Nocturnal rainfall amounts are higher with more

frequent trade-wind showers when trades are stronger because the convergence between the

decelerating incoming trades and the katabatic flow and the orographic lifting aloft (Chen and

Nash 1994; Yang and Chen 2007) are more significant when trades are stronger. In the

afternoon hours, orographic showers over land are more frequent with slightly higher rain rates for weak trade-wind days than strong trade-wind days. However, with relatively small sample sizes during HaRP, we would like to extend our analyses to the summers of 1997-2000.

4. Extending HaRP Analysis

Having only 12 strong and 12 weak composite days to work with from the 1990 HaRP

experiment, we extended our analyses with a larger database to provide more convincing

evidence of the impact of trade-wind speed on rainfall production during the diurnal cycle on the

windward side. Evolution of rainfall patterns during the diurnal cycle under different trade-wind

14 strength for the summer of 1997-2000 based on NCDC (National Climate Data Center) and

hydronet data will be examined in this section.

Daily rainfall totals indicate that stronger trade winds correspond to higher rainfall

amounts on the windward slopes of the island of Hawaii as the moisture laden trade-wind flow is

lifted over the windward slopes. The daily rain rate totals show a maximum over the windward

lowlands at 4 mm day-1 (Fig. 7a) for weak trade days, 8 mm day-1 for moderate trade days (Fig.

7b), 6 mm day-1 for strong trade days (Fig. 7c), 12 mm day-1 for moderately strong trade days

(Fig. 7d), and a dramatic 20 mm day-1 for very strong trade days (Fig. 7e). The daily rainfall

totals are greater when trades are stronger. These results are consistent with Carbone et al. (1998)

and Tuttle et al (2002). In addition to daily rainfall totals, we will investigate the impact of

trade-wind strength on rainfall distribution and amounts for different diurnal rainfall regimes.

Around midnight (2300-0300 HST), a northwest-southeast rainfall axis of 1.0 mm h-1 exists over the windward lowlands and the coast for very strong trade-wind days (Fig. 8a). In comparison, weak trade-wind days display a rainfall maximum rate of only 0.2 mm h-1 (Fig. 8b) located closer to the coastline. As suggested by Frye and Chen (2001) in a case study, the convergence between the katabatic winds and incoming decelerating trade-wind flow and the orographic lifting aloft are more pronounced under stronger trades, evidenced by having a rainfall axis with rain rates five times greater (Fig. 8a) than that of the weak trade-wind days (Fig.

8b).

The period before sunrise (0300-0700 HST) also displays a large disparity in rainfall rates. Similar to a strong (~ 11 m s-1) trade-wind case (Frye and Chen 2001), the maximum

rainfall axis remains over land for the very strong trade wind regime (Fig. 9a). Rainfall rates (1.2

mm h-1) near the coast to the northwest of the Hilo Bay area (Fig. 9a) indicate that the

15 convergence is stronger under stronger trade conditions, thereby resulting in more rainfall.

Precipitation for the weak trades (Fig. 9b) shows a lower rainfall rate and a shifting of the

maximum rainfall axis farther out toward the coastal region. It is apparent that under weaker

trade-wind conditions, the thermally induced katabatic flow extends farther offshore (Feng and

Chen 1998).

In the morning (0700-1100 HST), the axis of the rainfall maximum for the very strong

trade-wind category remains over the windward lowlands west of Hilo (Fig. 10a). With weak

trades, the axis is located farther eastward toward the eastern tip of the island of Hawaii (Fig. 10b)

showing the least amount of rainfall over land under this trade-wind regime. These results are

attributed to the fact that the katabatic flow is more pronounced under weaker trade-wind

conditions (Chan and Nash 1994; Feng and Chen 1998). Note that at 0900 HST, the splitting

airflow in the Hilo Bay area has an onshore wind component for strong trade-wind days during

HaRP, whereas for weak trade wind days, it still has an offshore component at this time (Chen

and Nash 1994; fig. 16c).

The period around noontime (1100-1500 HST) is characterized by the development of

upslope flow. For very strong trade winds, the rainfall axis has shifted from the lowlands in the early morning to slightly higher elevation (Fig. 11a). With weak trade winds, rainfall rates over

the windward lower slopes gradually increase as a result of the development of orographic clouds and showers (Fig. 11b). The afternoon period until sunset (1500-1900 HST) reveals that

under very strong trades, a rainfall axis of 0.6 mm h-1 exists over the windward lowlands (Fig.

12a). Under the weak trade-wind regime, rainfall rates over the lower windward slopes west of

Hilo (Fig. 12b) are higher than earlier due to orographic lifting provided by the combined

anabatic/trade-wind flow along the windward slopes. During HaRP orographic rainshowers on

16 the windward slopes (Station 11, elevation 1579 m) in the afternoon hours are more frequent (~

40%) for weak trades (Fig. 6c) than strong trades (Fig. 5c). Since there are no NCDC or

Hydronet stations located above the 1000-m terrain contour west of Hilo (Fig. 3), the maximum rainfall axis may be located at a higher elevation, similar to the situation for weak trade-wind conditions during HaRP (Fig. 6b). Under weak trades, the diurnal rainfall variations have pronounced afternoon maxima on the windward slopes.

For very strong trade-wind days, the rainfall rates over the windward lowlands in the early evening (1900-2300 HST) (Fig. 13 a) are higher than in the later afternoon hours (Fig. 12a).

The rainfall axis of 0.8 mm h-1 remains over the windward lowlands and extends to the west of the Hilo Bay area. For weak trade winds, there is a rapid decrease of rainfall on the windward slopes after sunset (Fig. 13b) as the daytime upslope wind is replaced by nighttime katabatic flow. Under stronger trades, in addition to upstream flow deceleration, focusing and enhancement of trade-wind showers in the convergence zone during the evening transition with a well-defined rainfall axis in the windward lowlands (Fig. 13a) (Frye and Chen 2001) is more pronounced. These results are consistent with those presented in Section 3 using a limited sample size during HaRP. The nocturnal rainfall is significantly higher and farther inland when trades are stronger. On the other hand, weak trade-wind days have a pronounced diurnal rainfall maximum on the windward slopes in the late afternoon hours. Throughout the diurnal cycle, rainfall amounts on the windward side are statistically higher with more frequent rainshowers under stronger trades, except in the late afternoon hours.

The daily rainfall totals show a maximum along the Kona lee-side coast at 2 mm day-1

(Fig. 7a) for weak trade days, 4 mm day-1 for moderate trade days (Fig. 7b), 6 mm day-1 for strong trade days (Fig. 7c), 8 mm day-1 for moderately strong trade days (Fig. 7d), and 12 mm

17 day-1 for very strong trade days (Fig. 7e). These results are consistent with Yang and Chen

(2003) (their fig. 18). The rainfall along the Kona coast has an early evening maximum (Chen and Nash 2004; Yang and Chen 2003). Yang and Chen (2003) and Yang and Chen (2007) suggest that the evening rainfall maximum there is caused by the convergence between the land breezes and the dynamically induced westerly flow between the two counter-rotating vortices in the lee (Smith and Grubišić, 1993). The westerly flow offshore and the convergence along the coast are expected to be more pronounced when trades are stronger (Yang and Chen 2003).

5. Summary

Rainfall rates on the windward side, except in the late afternoon hours, are statistically

higher throughout the diurnal cycle when trades are stronger. The thermally driven diurnal

winds distinctly affect the timing and location of rainfall occurrences under different trade-wind

strength. For normal or strong trade-wind regimes, the rainfall distribution has a nocturnal

maximum on the windward lowlands west of Hilo caused by the convergence of the katabatic

flow and the incoming decelerating trade-wind flow, and orographic lifting aloft. The maximum

rainfall axis shifts farther inland when trades are stronger.

For the weak (≤ 5 m s-1) trade-wind regime, the rainfall distribution has a late afternoon

maximum on the windward slopes as a result of the development of anabatic winds. In the

afternoon hours, the thermally driven anabatic winds are more significant when trades are

weaker. At night, under weaker trades, in addition to weaker orographic lifting aloft, the

convergence between the katabatic/offshore flow and the weaker incoming trade-wind flow

occurs farther eastward toward the coast and extends farther offshore in the early morning. As a

result, after midnight and in the early morning, rainfall is lower and the maximum rainfall axis is

displaced farther eastward toward the coast when trades are weaker.

18 REFERENCES

Austin, G. R., R. M. Rauber, H. T. Ochs III, and L. J. Miller, 1996: Trade-wind clouds and

Hawaiian rainbands. Mon. Wea. Rev., 124, 2126-2151.

Carbone, R. E., W. A. Cooper, W.-C. Lee, 1995: Forcing of flow reversal along the windward

slopes of Hawaii. Mon. Wea. Rev., 123, 3466-3480.

______J. D. Tuttle, W. A. Cooper, V. Grubišić, W. C. Lee, 1998: Trade-wind rainfall near the

windward coast of Hawaii. Mon. Wea. Rev., 126, 2847-2863.

Chen, F., and J. Dudhia, 2001: Coupling an advanced land surface-hydrology model with the

Penn State-NCAR MM5 modeling system. Part I: Model Implementation and Sensitivity.

Mon. Wea. Rev., 129, 569–585.

Chen, Y.-L., and J. Feng, 2001: Numerical simulations of airflow and cloud distributions over

the windward side of the island of Hawaii. Part I: The effects of trade wind inversion,

Mon. Wea. Rev., 129, 1117-1134.

______, and ______, 1995: The influences of inversion height on precipitation and airflow

over the Island of Hawaii. Mon. Wea. Rev., 123, 1660-1676.

______, and J. J. Wang, 1995: The effects of precipitation on the surface temperature and

airflow over the island of Hawaii. Mon. Wea. Rev., 123, 681-694.

______, and ______, 1994: Diurnal variations of surface thermodynamic fields on the

island of Hawaii. Mon. Wea. Rev., 122, 2125-2138.

______, and A. J. Nash, 1994: Diurnal variations of surface airflow and rainfall frequencies

on the island of Hawaii. Mon. Wea. Rev., 122, 34-56.

19 Dudhia, J. 1993: A nonhydrostatic version of the Penn state-NCAR mesoscale model: validation

tests and simulation of an Atlantic and cold front. Mon. Wea. Rev., 121, 1493-

1513.

Esteban, M. A., 2002: Impact of trade-wind strength on precipitation and circulation of the island

of Hawaii. MS Thesis, Department of Meteorology, University of Hawaii, Honolulu, HI

96822.

Feng, J., and Y.-L. Chen, 1998: Evolution of katabatic flow on the Island of Hawaii on 10

August 1990. Mon. Wea. Rev. 126, 2185-2199.

______, and ______, 2001: Numerical simulations of airflow and cloud distributions over

the windward side of the island of Hawaii. Part II: Nocturnal flow regime. Mon. Wea.

Rev., 129, 1135-1147.

Frye, J. L., and Y.-L. Chen, 2001: Evolution of downslope flow under strong opposing trade-

winds and frequent trade-wind rainshowers over the Island of Hawaii. Mon. Wea. Rev.,

129, 956-977.

Garrett, A. J., 1980: Orographic cloud over the eastern slopes of Mauna Loa Volcano, Hawaii,

related to insolation and wind. Mon. Wea. Rev., 108, 931-941.

Larson, R. N., 1978: Summer trade-wind rainfall in the Hawaiian Islands. M.S. Thesis,

Department of Meteorology, University of Hawaii, 84 pp.

Lavoie, R. L., 1967: Air motions over the windward coast of the Island of Hawaii. Tellus, 19,

354-358.

Leopold, L. B., 1949: The interaction of trade-wind and sea breeze, Hawaii. J. Meteor., 8, 533-

541.

20 Li, J., and Y.-L. Chen, 1999: A case study of nocturnal rain showers over the windward coastal

region of the Island of Hawaii. Mon. Wea. Rev., 127, 2674-2692.

Lin, Y.-L., T.-A. Wang, and R. P. Weglarz, 1993: Interactions between gravity waves and cold

air outflows in a stably stratified uniform flow. J. Atmos. Sci., 50, 3790-3816.

Rasmussen, R. M, and P. K. Smolarkiewicz, 1993: On the dynamics of Hawaiian cloud Bands.

Part III: Local aspects. J. Atmos. Sci., 50, 1560-1572.

______, ______, and J. Warner, 1989: On the dynamics of Hawaiian cloud bands:

Comparison of model Results with observations and island climatology. J. Atmos. Sci.,

46, 1589-1608.

Raymond, D. and R. Rotunno, 1989: Response of a stably stratified to cooling. J. Atmos. Sci.,

46, 2830-2837.

Sanderson, M., Ed., 1993. Prevailing Trade-winds, Weather and Climate in Hawaii. University

of Hawaii Press, 126pp.

Schär, C., and R. B. Smith, 1993: Shallow-water flow past isolated topography. Part I: Vorticity

production and wake formation. J. Atmos. Sci., 50, 1373–1400.

Schroeder, T. A., B. J. Kilonsky, and B. N. Meisner, 1977: Diurnal variation in rainfall and

cloudiness. UHMET Rep. 77-03, 67 pp. [Available from Department of Meteorology,

University of Hawaii, Honolulu, HI 96822.]

Smith, R. B., and V. Grubišić, 1993: Aerial observation of Hawaii’s wake, J. Atmos. Sci., 50,

3728-3750.

Smolarkiewicz, P., R. M. Rasmussen, T. L. Clark, 1988: On the dynamics of Hawaiian cloud

bands: Island forcing. J. Atmos. Sci., 45, 1872-1905.

Takahashi, T., 1981: Warm rain study in Hawaii – Rain initiation. J. Atmos. Sci., 38, 347-369.

21 Tuttle, J. D., R. E. Carbone, and J. J. Wang, 2000: Trade-wind rainfall near the windward coast

of Hawaii: Corrected data yield improved results. Mon. Wea. Rev. 128, 896-900.

Wang, J.-J., R. M. Rauber, H. T. Ochs III, R. E. Carbone, 2000: The effects of the island of

Hawaii on offshore rainband evolution. Mon. Wea. Rev. 128, 1052-1069.

______, and Y.-L. Chen, 1998: A case study of trade-wind rainbands and their interaction with

the island-induced airflow. Mon. Wea. Rev. 126, 409-423.

______, and ______, 1995: Characteristics of near-surface winds and thermal profiles on

the windward slopes of the Island of Hawaii. Mon. Wea. Rev. 123, 3481-3501.

Weiss, N.A., and M. J. Hassett, 1991: Introductory Statistics. Addison-Wesley Publishing Co.,

960 pages.

Yang, Y., and Y.-L. Chen, 2007: Effects of terrain heights and sizes on island-scale circulations

and rainfall. Mon. Wea. Rev., (In Press).

______, ______, and F. M. Fujioka, 2005: Numerical simulations of the island induced

circulations for the island of Hawaii during HaRP. Mon. Wea. Rev., 133, 3693-3713.

______, and ______, 2003: Circulations and rainfall on the leeside of the island of Hawaii

during HaRP. Mon. Wea. Rev., 131, 2525-2542.

Zhang, Y., Y. L. Chen, S. Y. Hong, H. M. H. Juang and K. Kodama, 2005: Validation of the

coupled NCEP Mesoscale Spectral Model and an advanced Land Surface Model over the

Hawaiian Islands: Part I: Summer trade-wind conditions over Oahu and heavy rainfall

events. Wea. Forecasting, 20, 847-872.

22 TABLE CAPTIONS

Table 1. Listing of different wind categories with respective speed range based on the

mean and standard deviation winds (m s-1) from HaRP, mean winds, estimated

Froude numbers and number of days.

Table 2. Dates used for NCEP/NCAR reanalysis wind data for each year from 1997-2000

under each wind category.

Table 3. Mean and standard deviation of inversion base heights under different

trade-wind regimes.

Table 4. National Weather Service hydronet stations over the island of Hawaii.

Table 5. National Climatic Data Center rainfall stations over the island of Hawaii.

23 FIGURE CAPTIONS

Figure 1. Locations of PAM stations during HaRP. Figure 2a. Composite sea-level pressure (hPa) and winds (m s-1) for very strong trade-wind days during 1997-2000. Figure 2b. Same as Figure 2a but for the weakest trade-wind days during 1997-2000. Figure 3a. Locations of National Weather Service Hydronet Stations. Figure 3b. Locations of National Climatic Data Center Stations.

Figure 4a. Averaged winds and virtual temperature perturbation (ΔTv) every 1 K between upstream and PAM data for transect from upper slopes to the coast of 45 HaRP days. Winds with one pennant, full barb, and half barb represent 5.0, 1.0, and 0.5 m s-1 respectively. Figure 4b. Rainfall rate every 0.4 mm h-1 for transect of windward coastal PAM stations of 45 HaRP days. Figure 5a. Same as Figure 4a but for 12 strong trade-wind days. Figure 5b. Same as Figure 4b but for 12 strong trade-wind days. Figure 5c. Rainfall incidence (%) of strong trade-wind days. Contour intervals every 20%. Figure 6a. Same as Figure 4a but for 12 weak trade-wind days. Figure 6b. Same as Figure 4b but for 12 weak trade-wind days. Figure 6c. Same as Figure 5c but for 12 weak trade-wind days. Figure 7a. Daily rainfall totals (mm day-1) for weak trade-wind days. Terrain contours every 1 km. Figure 7b. Daily rainfall totals (mm day-1) for moderate trade-wind days. Terrain contours every 1 km. Figure 7c. Daily rainfall totals (mm day-1) for strong trade-wind days. Figure 7d. Daily rainfall totals (mm day-1) for moderately strong trade-wind days. Figure 7e. Daily rainfall totals (mm day-1) for very strong trade-wind days. Figure 8. The rainfall rate (mm h-1) for midnight (2300-0300 HST) for a) very Strong, and b) weak trade-wind regimes in 1997-2000. Figure 9. Same as Figure 8 but for 0300-0700 HST. Figure 10. Same as Figure 8 but for 0700-1100 HST.

24 Figure 11. Same as Figure 8 but for 1100-1500 HST. Figure 12. Same as Figure 8 but for 1500-1900 HST. Figure 13. Same as Figure 8 but for 1900-2300 HST.

25 Table 1. Listing of different wind categories with respective speed range based on the mean and standard deviation winds (m s-1) from HaRP, mean winds, estimated Froude numbers and number of days. Speed Mean wind Froude Number of Wind Category (m s-1) (uˆ ) number Days

HaRP Strong 6.8 6.7 0.27 12

HaRP Weak 5.3 5.2 0.21 12

Very Strong >8.2 8.9 >0.3 39

Moderately 7.5-8.2 7.8 ~0.3 33 Strong

Strong 6.1-7.5 6.8 ~0.3 53

Moderate 5.6-6.1 5.9 ~0.2 17

Weak 4.8-5.6 5.2 ~0.2 20

Weakest <4.8 3.9 <0.2 11

Table 2. Dates used for NCEP/NCAR reanalysis wind data for each year from 1997-2000 under each wind category. NCEP/NCAR Reanalysis Wind Data Very Strong Days Moderately Strong Days 1997 1998 1999 2000 1997 1998 1999 2000 0723 0712 0711 0715 0716 0728 0713 0712 0801 0713 0712 0717 0718 0803 0716 0716 0802 0714 0721 0718 0722 0819 0720 0719 0722 0722 0725 0724 0723 0720 0723 0726 0726 0730 0725 0724 0804 0727 0818 0731 0807 0804 0805 0728 0819 0803 0822 0805 0820 0729 0820 0806 0824 0806 0821 0804 0821 0823 0808 0822 0805 0822 0809 0823 0806 0823 0810 0824 0813 0824 0811 0817

Strong Days Moderate Days 1997 1998 1999 2000 1997 1998 1999 2000 0711 0711 0714 0711 0725 0719 0809 0731 0712 0715 0715 0713 0726 0724 0811 0813 0717 0716 0717 0714 0727 0726 0721 0717 0718 0721 0808 0730 0729 0718 0719 0722 0809 0815 0804 0721 0724 0723 0810 0805 0727 0730 0727 0815 0807 0729 0803 0728 0821 0820 0806 0808 0729 0822 0807 0812 0730 0824 0813 0818 0801 0814 0819 0802 0818 0821 0803 0807 0812 0816

Weak Days Weakest Days 1997 1998 1999 2000 1997 1998 1999 2000 0715 0720 0731 0814 0713 0801 0719 0725 0802 0815 0714 0808 0720 0731 0810 0812 0809 0728 0802 0820 0813 0810 0811 0812 0817 0811 0814 0816 0817 0816 0819

Table 3. Mean and standard deviation of inversion base heights (hPa) under different trade wind regimes. Inversion Base Height (hPa) Trade Wind Regimes Mean St. Dev. 1990 Strong 779 61.1 1990 Weak 762 57.6 Very Strong 789 50.1 Moderately Strong 790 45.2 Strong 802 52.6 Moderate 793 48.8 Weak 808 47.8 Weakest 828 50.7

Table 4. National Weather Service hydronet stations over the island of Hawaii. Hydronet Latitude Longitude Elevation COOPID Station Name (oN) (oW) (m) 80 Laupahoehoe 19.983 155.233 109.7 81 Mountain View 19.550 155.167 466.3 82 Waiaha 19.633 155.950 469.4 83 Pahoa 19.550 155.983 149.5 84 Kealakekua 19.517 155.917 536.4 85 Pahala 19.200 155.483 256.0 86 Kamuela 20.050 155.650 877.8 87 Honokaa 20.067 155.467 387.1 88 Upolu AP 20.267 155.867 29.3 90 Waikii 19.864 155.654 1414.3 91 Piihonua 18.717 155.133 262.1 92 Waiakea Uka 19.667 155.133 304.8 94 Glenwood 19.517 155.167 798.6 95 Honaunau 19.467 155.883 414.6 96 Kahua Ranch 20.133 155.800 987.6 97 Kamuela Upper 20.033 155.667 926.6 98 Hilo 19.717 155.050 2.7 99 Kapapala Ranch 19.283 155.450 652.3 Table 5. National Climatic Data Center rainfall stations over the island of Hawaii.

NCDC Latitude Longitude Elevation COOPID Station Name (oN) (oW) (m)

Hawaii Volcanoes 511303 19.433 155.267 1210.4 National Park HQ 511339 Hawi 19.550 155.833 176.8 Hawaiian Ocean 511385 19.117 155.783 883.9 View Estate 511492 Hilo 19.717 155.050 9.1 512156 Huehue 19.750 155.967 597.4 512595 Kahuna Falls 19.867 155.150 371.9 512600 Kahua Ranch HQ 20.133 155.800 987.6 513072 Kamuela 20.050 155.650 877.8 513510 Kaumana 19.683 155.150 359.7 513925 Keaiwa Camp 19.233 155.483 518.2 513987 Kealakekua 4 19.517 155.917 432.8 514098 Keanakolu Camp 19.917 155.350 1609.3 515260 Lalamilo F O 20.017 155.683 797.1 515427 Laupahoehoe 19.983 155.233 115.8 516546 Mountain View #3 19.533 155.133 583.7 517209 Paauhau Mauka 20.067 155.450 341.4 517465 Pahoa School Site 19.500 155.950 208.2 518063 Pohakuloa 19.750 155.533 1984.6 518555 Puu Waa 19.783 155.850 768.1

Figure 1. Locations of PAM stations during HaRP.

1

) for very strong trade winds during 1997-2000. -1 a Figure 2a. Composite sea level pressure (hPa) and winds (m s

2

b 3

Figure 2b. Composite sea level pressure (hPa) and winds (m s-1) for weakest trade winds during 1997-2000.

a

Figure 3a. Locations of National Weather Service hydronet stations.

b

Figure 3b. Locations of National Climatic Data Center stations.

4

a

Figure 4a. Averaged winds and virtual temperature perturbation (ΔTv) every 1 K between upstream and PAM data for transect from upper slopes to the coast of 45 HaRP days. Winds with one pennant, full barb, and half barb represent 5.0, 1.0, and 0.5 m s-1 respectively.

5

b

Figure 4b. Rainfall rate every 0.4 mm h-1 for transect of windward coastal PAM stations of 45 HaRP days.

6

a

Figure 5a. Same as Figure 4a but for 12 strong trade-wind days.

7

bb

Figure 5b. Same as Figure 4b but for 12 strong trade-wind days.

8

c

Figure 5c. Rainfall incidence (%) of strong trade-wind days. Contour intervals every 20%.

9

a

Figure 6a. Same as Figure 4a but for 12 weak trade-wind days.

10

b

Figure 6b. Same as Figure 4b but for 12 weak trade-wind days.

11

c

Figure 6c. Same as Figure 5c but for 12 weak trade-wind days.

12

a

Figure 7a. Daily rainfall totals (mm day-1) for weak trade-wind days. Terrain contours every 1 km.

13

b

Figure 7b. Daily rainfall totals (mm day-1) for moderate trade-wind days. Terrain contours every 1 km.

14

c

Figure 7c. Daily rainfall totals (mm day-1) for strong trade-wind days.

15

d

Figure 7d. Daily rainfall totals (mm day-1) for moderately strong trade-wind days.

16

e

Figure 7e. Daily rainfall totals (mm day-1) for very strong trade-wind days.

17

a)

b)

Figure 8. The rainfall rate (mm h-1) for midnight (2300-0300 HST) for a) very strong and b) weak trade-wind regimes in 1997-2000.

18

a)

b)

Figure 9. Same as Figure 8 but for 0300-0700 HST.

19

a)

b)

Figure 10. Same as Figure 8 but for 0700-1100 HST.

20

a)

b)

Figure 11. Same as Figure 8 but for 1100-1500 HST.

21

a)

b)

Figure 12. Same as Figure 8 but for 1500-1900 HST.

22

a

b

Figure 13. Same as Figure 8 but for 1900-2300 HST.

23